中文摘要：

聚类分析是数据挖掘中的一个重要研究内容。按照数据对象间的关系进行聚类在许多情况具有特殊的意义。提出一种相容关系数据对象的聚类算法。该算法首先对每个数据对象按字典排序,利用相容集的反单调性性质来产生极大相容簇,即通过相容集的连接产生更高层的相容集的候选,再通过剪枝的方法来得到更高层的相容集。该方法可以有效压缩算法的搜索空间,是现有相容关系聚类算法的有益改进和补充。

英文摘要：

Cluster analysis is an important part to study in data mining.It is particularly meaningful to execute clustering according to relations among data objects.The paper puts forward a compatible relation data object clustering algorithm.The algorithm firstly sorts all data objects lexicographically to generate bigger compatible clusters through the anti-monotone property of compatible sets.Then through prune method it obtains higher level compatible sets.The method can efficiently compress the searching space of the algorithm,so that it is a beneficial improvement and complement for present compatible relation clustering algorithms.